Elevated pCO2 affects behavioural patterns and mechano-sensation in predatory phantom midge larvae Chaoborus obscuripes

Aquatic acidification is a major consequence of fossil fuel combustion. In marine ecosystems it was shown, that increasing pCO2 levels significantly affect behavioural and sensory capacities in a diversity of species. This can result in altered predator and prey interactions and thereby change community structures. Just recently also CO2 dependent acidification of freshwater habitats has been shown. Also here, increased levels of pCO2 change organisms’ behaviour and sensory capacities. For example, the freshwater crustacean Daphnia’s ability to detect predators and accurately develop morphological defences was significantly reduced, rendering Daphnia more susceptible to predation. It was speculated that this may have cascading effects on freshwater food webs. However, for a comprehensive understanding of how increased levels of CO2 affect trophic interactions, it is also important to study how CO2 affects predators. We tested this using the dipeteran phantom midge larva Chaoborus obscuripes, which is a world-wide abundant inhabitant of freshwater impoundments. We monitored activity parameters, predation parameters, and predation rate. Chaoborus larvae are affected by increased levels of pCO2 as we observed an increase in undirected movements and at the same time, reduced sensory abilities to detect prey items. This is likely to affect the larvae’s energy budgets. Chaoborus is a central component of many freshwater food-webs. Therefore, CO2 effects on predator and prey levels will likely have consequences for community structures.

. In fact, pCO 2 in freshwater lakes world-wide ranges from 3.1-fold below to 16-fold above atmospheric pCO 2 , with a mean of ~1000 µatm in 2007 6,8,9 . Moreover, in freshwater systems pCO 2 is often not stable throughout the day and throughout the season 10 . Regardless, authors have discussed 7 , prognosticated 5 and shown 3 that also freshwaters acidify with ongoing fossil fuel combustion. It is further discussed that pCO 2 peak periods intensify under climate change scenarios 3,7 .
Elevation of environmental pCO 2 levels accompanied by changes in aquatic pH has detrimental effects on organism fitness. Ocean acidification not only affects calcifying organisms where it reduces calcification abilities and growth rates 1 but also affects development 11 , reproduction 12 , metabolic rate 13 , sensory abilities and behaviour in a range of non-calcifying species 14,15 . Especially, when sensory abilities are impeded, this can change species interactions as organisms are hampered in their ability to detect con-and heterospecifics. For example, sensory cues passing between predator and prey cannot be correctly interpreted and anti-predatory responses are often suppressed which may result in altered community dynamics. This has been displayed in a range of marine fish where pCO 2 dependent reductions in pH affect sensory abilities. As a result, this can impair the detection of predators [16][17][18] . Moreover, behaviour is affected rendering some prey more active, so that prey is more vulnerable to predators 18 .
Similar observations have been made in freshwater prey species. For example, pink salmon larvae Oncorhynchus nerka show alterations in olfactory responses and anti-predator behaviour towards elevated pCO 2 19 . Similarly, shelter seeking behaviour in crayfish is affected 20 . More explicitly, in the freshwater crustacean Daphnia (which is a keystone species as it has a disproportionally large effect on its natural environment as it links primary produces to higher trophic levels), it was shown that sensory abilities are impaired by elevated levels of pCO 2 3 . In two species (i.e. D. pulex and D. longicephala) the ability to sense predators and develop accurate morphological defences was hampered, which renders them more susceptible to predation. This was discussed to have far reaching effects for the ecosystem as an inadequate defence expression may have cascading effects on all trophic levels 3 . However, the increasing prey vulnerability is just one side of pCO 2 impacts on predator-prey systems. To our knowledge, possible effects of constantly elevated pCO 2 levels on freshwater predators and their predation rates have not been shown. To uncover this, we here investigated the effect of increased pCO 2 levels on one central predator preying on first level consumers. The phantom midge larvae Chaoborus (diptera) is a typical inhabitant of standing freshwater bodies world-wide 21 . While they serve as an important food source for higher trophic levels including many fish species, the larvae themselves prey on ciliates, copepods, and cladocerans like Daphnia 22,23 . If the predator is affected by elevated levels of pCO 2 , and predation effectivity is reduced the overall food web effects become less straight-forward.
Predation parameters and predation rate. We found that pCO 2 exposed Chaoborus made significantly fewer strikes at their prey ( Table 2, Fig. 2A). Larvae exposed to elevated levels of pCO 2 stroke on average ~0.7 fold less in comparison to the larvae of the control conditions. While pCO 2 exposed larvae performed  Fig. 2C). The predation rate was significantly reduced in CO 2 exposed larvae. Larvae exposed to elevated levels of pCO 2 consumed ~0.6 less prey; on average only 5.29 ± 2.69 (mean ± StD.) Daphnia, while control larvae consumed 9.00 ± 3.06 (mean ± StD.) Daphnia (Fig. 3).  Table 1.

Discussion
While there is a wealth of research focussing on the effect of ocean acidification on species interactions, only little is known about the effects of elevated pCO 2 levels in freshwater ecosystems. Up to now there are only a handful of publications investigating pCO 2 dependent effects in freshwater taxa [24][25][26] and community structures 27 . In line with these previous observations, we here observe that Chaoborus larvae exposed for 24 h to high levels of pCO 2 are significantly affected in their behavioural patterns. In these 4 th instar larvae we observe behavioural changes in form of increased activity levels accompanied with reduced predatory strikes that result in a reduction of predation rate.

Increased activity levels.
It is already well known, that elevated levels of pCO 2 alter behavioural patterns in a diversity of marine species (reviewed in 28 ). Similarly, some freshwater species showed changes in behaviour 19,29 while others did not 30 . Lepomis macrochirus showed increased swimming velocities 31 , and Oncorhynchus nerka was shown to reduce anxiety 19 , while Gasterosteus aculeatus showed decreased boldness and curiosity during pCO 2 elevated conditions 29 . Not only vertebrates are affected by pCO 2 also other invertebrates, e.g. the freshwater mussel Lampsilis siliquiidea shows a reduction of valve movement. Crayfish Procambarus clarkii similarly reduced overall activity 20 . Our data contribute to these observations showing that behaviour is also affected in other invertebrates like dipteran larvae. Chaoborus larvae exposed to increased pCO 2 levels increase their overall activity patterns resulting from an increased number of turns and twitches. Directed movements such as forward movements and cleaning patterns (i.e. brushes) or dodges away from conspecifics were not affected. A reason for these increased activity levels may be that larvae try to escape these unfavourable environmental conditions, but this has to be tested in future experiments.
Importantly, our results show that pCO 2 effects cannot be inferred from other species as increasing and decreasing activity levels are observed. It is quite plausible that such pCO 2 induced higher activity levels incur energetic costs and higher energy demands.
Reduced sensory abilities affect predation rate. We find that Chaoborus strike less when exposed to elevated pCO 2 conditions. However, if they strike the probability of prey capture and prey ingestion is not changed. This indicates that not prey handling but prey detection is impaired. Chaoborus detect their prey using mechano-sensation 23,32 , which when impaired could explain for the reduced number of strikes. In consequence, we observe that predation rate is significantly reduced, i.e. larvae catch less prey during the same time period. This negatively affects their energy budget, and in combination with the possibly higher energy demand, will have implications for the larvae's life history parameters, and could affect population growth rates causing changes in community structures. In addition, it is plausible that larvae become more visible for their own predators.
In deed, this may suggest that predation pressure on the prey organism Daphnia is reduced. Daphnia itself however, are also affected by pCO 2 as their ability to adequately develop defences is decreased and thereby become more prone to predation 3 . How this will change population dynamics will probably depend on who of the two partners is affected more.

Mode of action.
At present, the precise way how CO 2 mechanistically affects organisms is still controversial and there are several plausible hypotheses. For example, CO 2 especially at high concentrations can have narcotic effects on nervous system functionality and could either affect the whole nervous system or only parts that are especially sensitive, thereby disbalancing motor actions and sensory modalities 33,34 .
Another hypothesis focusses on a change in GABA A receptor functioning, where the inhibitory action of GABA is reversed and becomes excitatory 15 . This can result in an increased excitability of the overall nervous system and has the potential to lead to the larvae's hyperactivity 34 . In an experiment mimicking GABA A receptor malfunctioning with the help of the GABA A receptor antagonist gabazine on Danio rerio brains showed an increased spontaneous firing rate which induced epileptic-like neuronal activity 35 . Such neuronal activities stemming from neuronal hyperexcitability could on the behavioural level cause the larvae's increase in undirected movements. An alternative hypothesis, discusses changes to glycine receptor functioning 34 . Glycine receptors are the dominant inhibitory receptors in many organisms, coupled to an ion channel permeable for chloride ions and carbonate HCO 3 − , acting in a similar manner like the GABA A system. It thus, represents an additional www.nature.com/scientificreports www.nature.com/scientificreports/ explanation of our observations. Which of these hypotheses holds true needs to be subject in future investigations using dedicated strategies e.g. as suggested by 34 .

Conclusion
Predator -prey interactions are powerful drivers of community dynamics very often regulated via sensory cues passing between predators and prey 28 . As predator and prey, both gather information about the presence of the other, the effects of pCO 2 increase on predator-prey dynamics will strongly depend on which participant is more compromised. However, the effect of CO 2 on organismal behaviour is not straightforward but defined by the CO 2 mode of action which is probably determined by the evolutionary history of the explicit species.
There is strong evidence, that when predator -prey interactions are impeded by anthropogenic stressors such as CO 2 , this may destabilize food-webs and lead to changes in biodiversity.

Material and Methods
Animal cultures. Chaoborus larvae hatch from eggs deposited in freshwater and pupate into adult midges after processing through four larval stages that are increasing in body size. Due to the gape limitation of their catching basket, they are size selective in their prey choice, and the smaller instars feed on smaller prey items like ciliates, while the larger instars feed on copepods and cladocerans like D. pulex 23 . To rule out size selection effects, we choose 4 th instar larvae as a representative instar, as these have been well investigated for preying on D. pulex in the 2 nd juvenile instar [36][37][38] . This predator-prey system has been well established in the past [36][37][38] . We anticipate that the results of this instar are well transferable to the other instars. These instars have the same predator capabilities, with the only exception that they prey on smaller items.
All experiments were conducted between September and December of 2018. Chaoborus obscuripes larvae of the 4 th juvenile instar were caught in the ponds of the Ruhr University's botanical gardens maximally 5 days prior to the experiments. During this season the ponds have a depth-dependent temperature range of 4 °C to 17 °C. During the summer, when larvae are most active, temperatures can reach up to 25 °C. To acclimate larvae to laboratory conditions, we gradually increased temperature by transferring the larvae from 4 °C via 15 °C to 22 °C in temperature-controlled rooms.
In detail, larvae were isolated from the ponds and twenty individuals were transferred into 1.5 L glass beakers (WECK, Germany) filled with artificial M4 media ((pH 8.0, with a pCO 2 of ~1,200 µatm, at 4 °C) see Table 3 39 ), and fed with 50 D. pulex juveniles daily. Larvae were first transferred to a cold room at 4 °C ± 1.0 °C for 24 h (16:8 day:night cycle). Subsequently, they were transferred to a room of 15 °C ± 1.0 °C for 48 h, where the medium warmed gradually to carefully acclimate the larvae. They were then transferred to a climatized laboratory set to 22 °C ± 1.0 °C again for gradual acclimation for 48 h. Larvae were not fed 24 h prior to the experiment.
As prey, we used age-synchronized D. pulex (also collected from the botanical gardens, but had been in the department's animal culture already since 2017). Daphnia were also kept in 1 L beakers in M4 at 20 °C ± 0.1 °C (16:8 day:night cycle) in densities of 30 animals per litre. D. pulex were fed every 48 h with the green algae Acutodesmus obliquus. Beakers were cleaned and water was exchanged on a weekly basis. To match 4 th instar Chaoborus larvae's prey spectrum, all experiments were conducted with D. pulex that had reached the second juvenile instar 22,38 . pco 2 conditions and experimental set-up. We set control conditions to a pCO 2 of ~1,300 μatm (Table 4) with a pH of ~8.0 and elevated pCO 2 conditions ~12,000 µatm (Table 4) with a pH of ~6.6 as published earlier 3 . These, in comparison to the ocean, high values in the control condition were selected based on the global mean pCO 2 in freshwater habitats 9 . Similarly, we selected the treatment condition of ~12,000 µatm based Figure 3. Chaoborus predation rate. Chaoborus larvae exposed to increased levels of pCO 2 consumed significantly less D. pulex. Statistics displayed in Table 5.  www.nature.com/scientificreports www.nature.com/scientificreports/ on currently observed pCO 2 maxima of ~10,000 µatm, resulting from the diel and seasonal fluctuations 40 . The elevated pCO 2 condition was achieved via bubbling and setting the pH to 6.6 prior to the experiments using pH and temperature probes (by Aqua Medic, Germany), documenting temperature levels alongside being stable at ~22 °C. 200 mL of all media were titrated using a Titrino (Methrohm, Switzerland) after the experiments to validate pCO 2 and temperature conditions. We determined temperature, pH as well as acid and base capacity for pCO 2 calculation via Phreeqc 3,41 (see Table 4). The control and the elevated pCO 2 condition were both tested on the same day but consecutively. To rule out day-time and circadian rhythm dependent effects, we randomized the sequence in which the two treatments were measured. Each experimental trial started between 9 and 10 a.m. for the first condition and between 12 and 1 p.m. for the alternative condition with the exposure of three Chaoborus larvae to control and three Chaoborus larvae to elevated pCO 2 conditions for 24 h in custom made water tanks (12.5 cm × 2 cm × 10.5 cm). Tanks were covered airtight by sealing the lid with parafilm to prevent outgassing. All experiments were performed at a constant temperature (see Table 4) in a temperature-controlled room in above mentioned water tanks. On the following day, i.e. 24 h post exposure (i.e. between 9 and 10 a.m. and between 12 and 1 p.m.), the experiments started with the addition of 100 second juvenile instar D. pulex. Predator and prey were allowed to acclimate for 10 min. Subsequently, larvae predation parameters were monitored for 1 h. During this monitoring period we additionally recorded 5 film sequences of 10 min using an iPhone 7 (Mac iOS 12.4.2 Apple Inc.) interspaced by 2 min. breaks. For that the iphone was fixed in 13 cm distance from the tank using a tripod (KobraTech, Germany). Iphone camera orientation was positioned in parallel to the frontal plane of the tank. To ensure homogeneous illumination, a diffusor plate (customized translucent PVC plate) was positioned behind the tank illuminated by a 15 W LED lamp (IP 65, LE, Germany). As the larvae are about 1.7 to 2.0 cm in size, this allowed us to record activity patterns and predation parameters in the glass tanks over the experimental period. All experimental trials were replicated 17 times.
Analysis of activity patterns. We analysed activity patterns based on recorded videos. Sequences were viewed and analysed using iMovie (Mac OS Mojave Version 10.14.6, Apple inc.). The larvae display distinctive activity patterns, which we categorized into movement categories. A 'move' was defined as a forward movement of a larva. A 'turn' was defined as a 180° change in orientation, while a 'spin' was defined as a full 360° turn around the body axis. A 'twitch' was defined as a sudden, undirected convulsive movements. A 'dodge' describes the movement, when larvae tried to avoid contact to other larvae. A brush describes a movement in which the larvae clean their tail fan. The category total activity level comprises the sum of all movement categories of the experimental population.
Predation parameters. During the one hour observation period, we counted all strikes, catches, and ingestions of the larval attacks (according to 22 ) and thereby determined the population's predation parameters. We then calculated the proportion of strikes that led to catches (in %) and the proportion of catches (%) during this one hour.
Predation rate. To analyse the effects of pCO 2 on the predation rate of Chaoborus, we reared one Chaoborus larvae for 24 h in 250 mL M4 either in the control condition or aerated with CO 2 (  Statistics. In total, we performed 17 experimental replicates in the control condition and 17 experimental replicates in the pCO 2 condition. In the pCO 2 condition one replicate had to be excluded due to instabilities in pCO 2 (therefore N control = 17; N pCO2 = 16). Activity patterns and predation parameters were calculated as the summated activity of all three larvae and therefore represent the population's total activity. We recorded activity parameters (i.e. total activity, turns, twitches, dodges, spins, moves and brushes) 5 times (for 10 min) within one hour observation time (N control = 85 and N pCO2 = 80). To determine if elevated pCO 2 has a significant effect on activity patterns we performed generalized linear mixed models (GLMMs) in combination with a poisson distribution for count data, where the different activity parameters were used as response variables, and treatment (control, elevated pCO 2 ) was used as fixed effect. As we measured 5 times per 1 h, we included time as a random factor (to reflect a repeated measures design). We fitted the GLMMs using the glmer function implemented in the lme4 package in R; www.raproject.org 42 ).
To analyse count data obtained in the predation parameter 'strike' and predation rate, we performed linear mixed models using the glm function and a poisson regression in R. Percent data (i.e. relative catches, and relative ingestions) were analysed using a beta regression using the betareg function in the Betareg package in R according to 43 . As relative ingestion data contained 0 and 1, data was transformed as suggested by 44 using formula x′ = (x(N − 1) + s)/N (with N = sample size and s = 0.5). All models were validated by visual inspection of the normalised residuals based on the REML fit against fitted values to identify possible violation of homogeneity, according to 45,46 . We tested for overdispersion; a dispersion value of <2 was considered not overdispersed 46 . None of our data was overdispersed.